During a discussion in our lab last month, a professor mentioned to me that the behavior of Self-Organizing Maps can be described in terms of repeated applications of the Moore-Penrose psuedoinverse, in a vaguely similar way to how single-layer neural networks with Hebbian learning can be described by Principle Component Analysis.
An engineering student friend of mine confirmed that the MP inverse is used for dimensionality reduction, but I can't find any material on this online. Can someone point me to an article, paper, or book on the subject?
